TFAP2C, TEAD4 and RHOA are sufficient to advance cell fate commitment

We previously showed that ectopic expression of TFAP2C, TEAD4 and RHOA (named TTRhoA hereafter) accelerates embryo polarization. Mouse embryos typically polarize at the eight-cell stage; however, embryos expressing TTRhoA polarize already at the four-cell stage28. Given that the potential of polarized outer cells to regenerate ICM is progressively lost as blastomeres become committed to the TE fate20, we wished to determine whether these three factors are also sufficient to accelerate TE commitment. To test this, we performed a blastocyst reconstruction assay (Fig. 1c). We microinjected both blastomeres of two-cell embryos with EZRIN–red fluorescent protein (RFP) messenger RNA (mRNA) (to mark the apical domain), with or without TTRhoA, using a technique that does not impair development and allows the embryo to develop beyond implantation31. Polarized blastomeres were sorted from nonpolarized blastomeres at the mid-16-cell stage, re-aggregated and allowed to develop to the blastocyst stage. We determined the proportion and number of ICM cells relative to total cells in reconstructed blastocysts (Fig. 1c–e and Extended Data Fig. 1a–f). Compared with controls, blastocysts derived from polarized blastomeres expressing TTRhoA had a lower number of ICM cells and therefore a lower ICM ratio (Fig. 1e and Extended Data Fig. 1b). These results indicate that TTRhoA overexpression not only accelerates the timing of embryo polarization28 but also accelerates commitment to the TE fate in blastomeres.

TFAP2C, TEAD4 and RHOA advance apical domain aging and inhibit Hippo signaling

Hippo signaling pathway components at the apical domain in the TE are inactive, leading to nuclear localization of unphosphorylated, active YAP7,14. In contrast, active Hippo signaling in the ICM generally triggers YAP phosphorylation (p-YAP), cytoplasmic retention and degradation32. Thus, the Hippo pathway negatively regulates YAP activity.

To determine how the acceleration of polarization and TE commitment in TTRhoA blastomeres affects Hippo signaling, we expressed the apical marker, EZRIN–RFP, with or without TTRhoA in one blastomere at the late two-cell stage. The uninjected blastomere served as a noninjection control (Fig. 2a). We found that none of the EZRIN–RFP blastomeres in control early eight-cell embryos displayed an apical domain3,5, whereas about 30% of the blastomeres in EZRIN–RFP + TTRhoA had an apical domain already at the early eight-cell stage28 (Fig. 2b,c and Extended Data Fig. 2).

Fig. 2: Tfap2c, Tead4 and activated RhoA coordinate Hippo inactivation with apical domain formation.

a, A schematic showing the workflow for experiments in b and c. b, Embryos injected with EZRIN–RFP only (as a control) or TTRhoA mRNAs were analyzed at the early eight-cell stage to reveal EZRIN–RFP and AMOT. The yellow squares indicate the magnified regions. The arrows indicate magnified cells. c, Quantifications of apical membrane enrichment of AMOT in cells expressing EZRIN–RFP or with TTRhoA. Data are shown as individual data points with box and whisker plots (lower: 25%; upper: 75%; line: median; and whiskers: min to max). Each dot indicates an analyzed cell. N = 12 cells for EZRIN–RFP and N = 8 cells for TTRhoA. N = 2 experiments. ***P = 0.0002, two-sided Mann–Whitney test. d, A schematic of TTRhoA overexpression for experiments shown in e–g. e, Embryos overexpressing EZRIN–RFP only (as a control) or TTRhoA, immunostained at mid eight-cell stage for DNA (DAPI), YAP and EZRIN–RFP. The pink arrows indicate apolar cells and yellow arrows indicate polar cells. Quantifications are shown in f. f, Quantification of the YAP N/C ratio in the polar or apolar cells of embryos overexpressing EZRIN–RFP only or TTRhoA. Data shown as individual data points with mean, cyan dots indicate polar cells and red dots indicate apolar cells. N = 12 embryos for EZRIN–RFP only and N = 29 embryos for the TTRhoA group, N = 4 experiments, ****P < 0.0001, two-way ANOVA test. YAP N/C ratios between polar and apolar cells are statistically different in the TTRhoA group but not in the EZRIN–RFP only group. g, Embryos overexpressing EZRIN–RFP only (as a control) or TTRhoA analyzed at mid eight-cell stage for EZRIN–RFP or p-YAP. The arrows indicate the apolar cells in TTRhoA overexpressing embryos. h, Quantification of the cytoplasmic ratio of p-YAP between the polar and apolar cells in embryos overexpressing of EZRIN–RFP only or TTRhoA. Data shown as individual data points with mean indicated by the line. N = 7 embryos for EZRIN–RFP only and N = 11 embryos for the TTRhoA group, N = 4 experiments and *P < 0.05, Mann–Whitney test. The lower cytoplasmic level of p-YAP in polar versus apolar cells in TTRhoA embryos versus controls. For all quantifications, data are shown as individual data points with mean. Scale bars, 15 μm.

Source data

To investigate the consequences of precocious apical domain formation on Hippo signaling, we wished to determine YAP localization in TTRhoA embryos versus controls. To this end, we upregulated expression of TT at the two-cell stage, introduced activated RHOA at the four-cell stage and examined embryos at the mid eight-cell stage (Fig. 2d). Both EZRIN–RFP-only and TTRhoA mid eight-cell stage embryos had a mixture of polarized and unpolarized cells, with more polarized cells in TTRhoA embryos than in controls (Fig. 2e–h). In control embryos, the polarized and unpolarized blastomeres displayed similar nuclear-to-cytoplasmic (N/C) ratios of YAP (Fig. 2e,f and Extended Data Fig. 3a). In contrast, in TTRhoA mid eight-cell embryos, the N/C ratios of YAP were higher in polar versus apolar blastomeres (Fig. 2g,h and Extended Data Fig. 3b). Moreover, the levels of cytoplasmic p-YAP were lower in polar versus apolar blastomeres of TTRhoA versus control embryos (Fig. 2h). We often detected nuclear p-YAP in blastomeres, which has been observed in other contexts, particularly in sparsely plated cells33,34 (Discussion). Overall, these data suggest that Hippo signaling diminishes after polarization of TTRhoA embryos, which indicates that TTRhoA not only advance polarization timing but also enhance the function of the apical domain, resulting in reduced Hippo signaling in blastomeres.

TFAP2c and TEAD4 activate TE and ICM genes to create a bipotent state

We previously found that TTRhoA embryos had precocious expression of the TE specifier Cdx2 (ref. 28). To determine whether this is related to advanced polarization, we microinjected EZRIN–RFP with or without TTRhoA in one blastomere of two-cell stage embryos (Extended Data Fig. 4a). These embryos express an endogenous CDX2–green fluorescent protein (GFP) reporter35, which showed low CDX2–GFP expression in blastomeres of control eight-cell embryos (Extended Data Fig. 4b) as expected36,37. TTRhoA overexpression led to an upregulation of CDX2–GFP not only in polarized cells, but also in apolar cells (Extended Data Fig. 4b,c). Consistent with a previous study22, Tfap2c depletion led to reduced Cdx2 levels in mid eight-cell stage embryos by RNA sequencing (Extended Data Fig. 4d). Moreover, depletion of both Tead4 and Tfap2c exacerbated the decrease in Cdx2 mRNA levels (Extended Data Fig. 4d). Analysis of CDX2–GFP reporter embryos revealed that depletion of Tfap2c and/or Tead4 in two-cell embryos significantly reduced the GFP levels in apolar blastomeres of eight-cell embryos, consistent with the RNA-sequencing data (Extended Data Fig. 4d–f). Thus, Tfap2c and Tead4 each promote expression of the TE marker Cdx2 even before lineage specification, in both polar and apolar cells.

The TE marker Gata3 is expressed independently of Cdx2 (ref. 15) and our RNA-sequencing data revealed that depletion of Tfap2c and Tead4 also reduced the expression of Gata3 in eight-cell embryos (Fig. 3a). Embryos expressing a GFP reporter driven by the Gata3 promoter38 displayed GFP expression in all cells at the early 16–32 cell stage, followed by preferential expression in outer polar cells at the 32–64 cell stage (Extended Data Fig. 5a,b) as expected15. Notably, depleting Tfap2c and Tead4 at the two-cell stage strongly abrogated Gata3 expression in both polar and apolar blastomeres at the early 16-cell stage (Fig. 3b–d). Moreover, GATA3 was prematurely expressed in TTRhoA embryos (2.7-fold upregulation at the late eight-cell stage; Fig. 3e,f). Notably, embryos expressing Tfap2c and Tead4 alone, as well as TTRhoA embryos treated with the Rho inhibitor C3-transferase39, showed polarity-independent induction of GATA3 at the eight-cell stage but not after (Fig. 3g,h). Thus, Tfap2c and Tead4 also each promote expression of the TE marker Gata3 even before lineage specification.

Fig. 3: Tfap2c and Tead4 regulate the expression of Gata3 before polarization.

a, RNA-sequencing analysis of Gata3 expression level at the eight-cell stage in embryos injected with dsGFP, dsTfap2c, dsTead4 or dsTfap2c + dsTead4. N = 5 samples for dsGFP and dsTfap2c + dsTead4 and N = 4 samples for dsTfap2c and dsTead4. Data are shown as mean ± s.e.m. *P < 0.05, Kruskal–Wallis test. N = 2 collections. b, A schematic of workflow for experiments in c–h. One blastomere of the two-cell stage embryo was injected with mRNA encoding Ezrin only (as a control), or also with dsRNA targeting Tfap2c and Tead4, or also with TTRhoA mRNA. c, Representative images of GATA3–GFP expression level in 8–16-cell stage embryos injected with the indicated dsRNA as described in b. Quantifications are shown in d. *P < 0.05, Mann–Whitney test. d, Quantification of the level of GFP in control and embryos injected with dsRNA targeting Tfap2c and Tead4. *P = 0.0262, Mann–Whitney test. Data are shown as mean ± s.e.m. N = 7 embryos for the Ezrin-only group and N = 7 embryos for the dsTfap2c + dsTead4 group. N = 2 experiments. e, Representative images of GATA3–GFP transgenic late eight-cell embryos, after injection with EZRIN only or TTRhoA, as described in b. The arrows indicate an injected cell. The number of embryos and quantifications shown in f. BF, bright field. f, Quantifications of normalized GATA3–GFP signal intensity in the indicated overexpression conditions. For normalization, GFP signal in injected cells were normalized against the noninjected cells. Data are shown as mean ± s.e.m. The numbers indicate the number of embryos analyzed. *P = 0.0218, one-way ANOVA test. g, Representative images of GATA3–GFP expression in embryos injected with EZRIN–RFP mRNA and TTRhoA, as indicated in b, and treated with water (control) or C3-transferase (RhoA inhibitor) at the late eight-cell stage. Quantifications are shown in h. h, A time course of the normalized GATA3–GFP signal intensity in cells overexpressing EZRIN–RFP only (control), or also exposed to TTRhoA, RhoA inhibitor or TTRhoA + RhoA inhibitor. Data are shown as mean ± s.e.m. n = 7 embryos for each group. The yellow region indicates the early stages of developmental when Gata3 expression is insensitive to RhoA activity (before the 16-cell stage) and the purple region indicates RhoA-sensitive stages (after the 16-cell stage). Scale bars, 15 μm.

Source data

We also found that Tfap2c leads to the upregulation of the ICM specifiers, Nanog and Oct4, which are expressed in all blastomeres at the eight-cell stage and become restricted to the ICM at the mid-blastocyst stage40,41,42,43,44. We found that the depletion of TFAP2C, but not TEAD4, led to downregulation of Oct4 (Fig. 4a–c), consistent with previous indications29,45,46. Moreover, we found that Nanog expression was downregulated upon the depletion of TFAP2C but not TEAD4 (Fig. 4a–c). In addition, co-expression of TFAP2C- and TEAD4-induced upregulation of Nanog and Oct4 at the eight-cell stage (Fig. 4d–g). These data further suggest that TFAP2C activates ICM genes in bipotent cells, in agreement with a recent study showing that TFAP2C binds and activates not only TE genes but also some early ICM genes12.

Fig. 4: Tfap2c and Tead4 regulate the expression of ICM specifiers before polarization.

a, The expression of Nanog, Pou5f1 (Oct4) and Klf5 by bulk RNA sequencing in the indicated conditions. **P = 0.0042 for Nanog, P = 0.0021 for Pou5f1 and P = 0.0019 for Klf5, Kruskal–Wallis test. N = 5 samples for dsGFP and dsTfap2c + dsTead4 and N = 4 samples for dsTfap2c and dsTead4. N = 2 collections. Data are shown as mean ± s.e.m. b, Representative images of embryos injected with Cas9 mRNA or with gRNAs targeting Tfap2c gene locus, to fix at the mid eight-cell stage and stain for NANOG and TFAP2C. The quantification is shown in c. c, Quantification of NANOG expression in Cas9-only (control) or Tfap2c-depleted cells by CRISPR–Cas9 shown in b. ****P < 0.0001, two-sided Student’s t-test. N = 27 embryos for the Cas9-only group and N = 10 embryos for Tfap2c KO embryos. N = 2 experiments. Data are shown as individual data points with box and whisker plots (bottom: 25%; upper: 75%; line: median; whiskers: min to max). d, Representative images of embryos injected with EZRIN–RFP mRNA alone or with Tfap2c and Tead4 mRNA, and visualized NANOG expression at the mid eight-cell stage. The embryos were injected at the two-cell stage and fixed at the mid eight-cell stage. e, Quantification of NANOG protein levels in conditions showing in d. N = 39 cells for the EZRIN–RFP group and N = 17 cells for the Tfap2c + Tead4 group. N = 2 experiments. ****P < 0.0001, Mann–Whitney test. Data are shown as individual data points with box and whisker plot (bottom: 25%; upper: 75%; line: median; whiskers: min to max). f, Representative images of embryos injected with EZRIN–RFP mRNA alone or with Tfap2c and Tead4 mRNA at one cell of the two-cell stage and fixed at the early to mid eight-cell stage, and visualized POU5F1 (OCT4) expression at the mid eight-cell stage. g, Quantification of OCT4 protein levels in cells from conditions shown in f. N = 28 cells for the Ezrin–RFP group and N = 15 cells for the Tfap2c + Tead4 group. N = 2 experiments and ****P < 0.0001, Mann–Whitney test. Data are shown as individual data points with box and whisker plots (bottom: 25%; upper: 75%; line: median; whiskers: min to max). Scale bars, 15 μm.

Source data

These data suggest temporally distinct functions for Tead4 and Tfap2c during development. Before lineage specification, Tfap2c and Tead4 promote biopotency through inducing the expression of ICM and TE markers and prime both fates. After specification, Tead4 and Tfap2c promote the TE fate20.

TFAP2c and TEAD4 promote the expression of Hippo regulators and Hippo signaling

We next wished to determine when Hippo signaling is initiated during development. To this end, we examined a previously published RNA-sequencing dataset47 to analyze the expression of Hippo signaling components. We found that Amot, Amotl2 and Lats2 were not expressed in the zygote but increased at least tenfold from the 2–4-cell stage to the blastocyst stage (Fig. 5a and Extended Data Fig. 6d), concomitant with increased expression of Tfap2c and Tead4 (ref. 28). To determine whether Tfap2c and Tead4 control the induction of these Hippo pathway components, we simultaneously depleted TFAP2C and TEAD4 by RNA interference (RNAi) and clustered regularly interspaced short palindromic repeats (CRISPR), which reduced the levels of all three transcripts at the eight-cell stage28 (Fig. 5b and Extended Data Fig. 5c).

Fig. 5: Tfap2c and Tead4 regulate Amot, Amotl2 and Lats2 and activate Hippo signaling.

a, The expression profile of Amot, Amotl2 and Lats2, data obtained from Deng et al.47b, The expression of Amot (***P = 0.0052, Kruskal–Wallis test), Amotl2 (*P = 0.0115, one-way ANOVA) and Lats2 (****P < 0.0001, Kruskal–Wallis test) by bulk RNA sequencing in the indicated conditions. The expression level is shown as mean ± s.e.m. N = 5 samples for dsGFP and dsTfap2c + dsTead4 and N = 4 samples for dsTfap2c and dsTead4. N = 2 collections. c, Late eight-cell embryos injected with EZRIN–RFP + dsGFP (control) or EZRIN–RFP + dsTfap2c in half embryo and immunostained AMOT, EZRIN–RFP and DNA (DAPI). BF, bright field. d, Quantification of plasma membrane-localized AMOT as in c. N = 17 cells for EZRIN–RFP and N = 14 cells for dsTfap2c. N = 2 experiments. **P = 0.0087, two-sided Mann–Whitney test. e, Mid-eight-cell embryos injected with Cas9 mRNA or with Tfap2c sgRNAs stained with TFAP2C and AMOT. f, Quantification of membrane AMOT as in e. N = 62 cells for Cas9 only and N = 18 cells for Tfpa2c CRISPR. N = 2 experiments. ****P < 0.0001, two-sided Mann–Whitney test. g, Late four-cell embryos overexpressing EZRIN–RFP or with Tfap2c in half embryo immunostained with AMOT, EZRIN–RFP and DNA (DAPI). h, Quantification of membrane AMOT as in k. N = 11 embryos for EZRIN and N = 9 embryos for the Tfap2c group. N = 2 experiments. **P = 0.0013, two-sided Mann–Whitney test. i, Late eight-cell embryos injected with dsRNA targeting GFP (control) or Tead4 dsRNA in half embryo and immunostained EZRIN–RFP and p-YAP. j, Quantification of cytoplasmic p-YAP levels as in g. ***P < 0.001, Mann–Whitney test. N = 8 embryos for dsGFP and N = 8 embryos for dsTead4. N = 2 experiments. k, Mid eight-cell embryos injected with Cas9 mRNA or with Tead4 sgRNAs stained with TEAD4 and p-YAP. l, Quantification of cytoplasmic p-YAP levels as in i. N = 87 cells for Cas9 and N = 68 cells for Tead4 CRISPR. N = 2 experiments. ****P < 0.0001, two-sided Mann–Whitney test. m, Mid eight-cell embryos overexpressing EZRIN–RFP or with Tead4 in half embryo immunostained EZRIN–RFP and p-YAP. n, Quantification of cytoplasmic p-YAP as in m. N = 10 embryos for EZRIN and N = 16 embryos for the Tead4 group. N = 2 experiments. ***P = 0.0002, two-sided Mann–Whitney t-test. For d, f, h, j, l and n, data are shown as individual data points with box and whisker plots (bottom: 25%; upper: 75%; line: median; whiskers: min to max). Scale bars, 15 μm.

Source data

We found that RNAi-mediated depletion of TFAP2C in zygotes reduced Amot expression by 79% in eight-cell embryos. To confirm this result, we also carried out CRISPR-mediated depletion of TFAP2C, which we previously showed led to efficient gene editing of the Tfap2c locus28. We found that both RNAi- and CRISPR-mediated depletion of TFAP2C reduced the levels of AMOT protein in late eight-cell embryos (Fig. 5c–f). In particular, the relative levels of active AMOT at the plasma membrane versus cytoplasm were reduced by 21.4% and 71% upon depletion of TFAP2C by RNAi and CRISPR8,16,28 (Fig. 5c–f). In contrast, overexpression of TFAP2C alone in two-cell embryos led to increased levels of active AMOT at the plasma membrane (Fig. 5g,h).

In addition, RNAi-mediated depletion of Tead4 in zygotes reduced the expression of Amotl2 and Lats2 by 97% and 83%, respectively (Fig. 5b). Consistent with reduced LATS2 activity and diminished Hippo signaling48, TEAD4 depletion by RNAi or CRISPR decreased the levels of cytoplasmic p-YAP by 46% and 52.8% in eight-cell embryos, respectively (Fig. 5i–l). Moreover, overexpression of Tead4 led to increased levels of p-YAP in the cytoplasm (Fig. 5m,n), consistent with elevated Hippo signaling. Yet we previously found that TEAD4 overexpression increases the N/C ratio of YAP before blastomere polarization28. In agreement with this, nuclear localization of active YAP has been shown in other contexts to initiate a negative feedback loop via Hippo-mediated phosphorylation of cytoplasmic YAP30. These results indicate that Tfap2c and Tead4 each positively regulate Hippo signaling by promoting the expression of Hippo pathway components before polarization, but also negatively regulate Hippo signaling by promoting apical domain formation and aging (Fig. 2). The overall result is intermediate Hippo signaling and the bipotent TE/ICM fate.

Before lineage specification, blastomeres express the transcription factor Kruppel-like factor 5 (KLF5), which directly induces both ICM and TE specification genes49. Although Klf5 was downregulated in blastomeres depleted for Tfap2c and Tead4 (Fig. 4a), it was not elevated along with the ICM and TE lineage specifiers in embryos with ectopic expression of TFAP2C and TEAD4 (Extended Data Fig. 6a–c). To determine whether Klf5 is required for the Tfap2c- and Tead4-dependent regulation of lineage markers, we depleted KLF5 in embryos overexpressing TFAP2C and TEAD4. RNAi-mediated depletion of KLF5 effectively eliminated KLF5 protein, but did not affect upregulation the TE marker CDX2 nor did it compromise the ICM feature of increased levels of active AMOT at the plasma membrane (Extended Data Fig. 6). These data suggest that Tfap2c and Tead4 regulate Klf5 and bipotency independently.

Our results overall show that Tfap2c promotes the expression of the TE regulators Cdx2 and Gata3, the ICM regulators Nanog and Oct4, and the Hippo regulator Amot in bipotent cells; Tead4 similarly promotes the expression of Cdx2 and Gata3 and the Hippo regulators Amotl2 and Lats2 in bipotent cells. As a consequence, bipotent blastomeres have a sum of intermediate Hippo signaling and display both active nuclear YAP and inactive cytoplasmic p-YAP.

Potential role of TFAP2C in the first cell fate decision in human embryos

Mammalian development at the preimplantation stage is evolutionarily conserved in both gross morphology and expression of the representative lineage markers50,51,52,53,54. To determine whether the gene expression network that we identified here in the mouse embryo is conserved in humans, we first examined published RNA-sequencing datasets. We found that the expression of TFAP2C increases from the four-cell stage whereas the expression of TEAD4 increases from the eight-cell stage in the human embryo55 (Fig. 6a). We confirmed this early presence of TFAP2C before cell polarization in human embryos by immunostaining (Fig. 6b). ATAC-sequencing data also suggest that genomic binding sites for TFAP2C, but not TEAD4, are highly accessible at the precompaction stage of the human embryo55.

Fig. 6: TFAP2C regulates gene expression prior to cell compaction in the early human embryo and the model.

a, The mRNA expression profiles of TFAP2C and TEAD4 in preimplantation human embryos. Data retrieved from Stirparo et al., 2018 (ref. 50). Data are shown as mean ± s.e.m. b, Human embryos before and after polarization were fixed and stained for PARD6 and TFAP2C. N = 4 embryos were examined. c–f, mRNA expression profile of Gata3/GATA3 (c), Amot/AMOT (d), Amotl2/AMOTL2 (e) and Lats2/LATS2 (f) in preimplantation mouse and human embryos. Data retrieved from Stirparo et al., 2018 (ref. 50). g, The University of California, Santa Cruz browser view showing accessible chromatin regions in Gata3/GATA3 and Amot/AMOT, Amotl2/AMOTL2 and Lats2/LATS2 loci in the mouse and human embryos at different stages, determined from ATAC-sequencing data. Mouse data retrieved from Wu et al.61. Human data retrieved from Wu et al.55. Scale bars, 15 μm.

Source data

When we compared the RNA-sequencing and ATAC-sequencing datasets between mouse and human embryos, we found that the TE transcription factor Gata3/GATA3 and Hippo components Amot/AMOT, Amotl2/AMOTL2 and Lats2/LATS2, which are regulated by Tead4 and Tfap2c in mouse embryos28 shared similar zygotic patterns of expression to that of the human embryo (Fig. 6c–g). We identified specific cis-regions near these genes that become accessible only around the four- or eight-cell stage in the human embryo. These cis-regions contained both TFAP2C and TEAD4 consensus sequences in the mouse genome, but only TFAP2C consensus sequences were detected in the human genome (Extended Data Fig. 7). It will be of interest to determine whether TFAP2C regulates bipotency and a bistable switch in blastomeres of the human embryo, in addition to its known roles in polarization and lineage segregation.